-1
$\begingroup$

Does this statement correct? if it does how we can prove it. In Banach spaces a map is local diffeomorphism if and only if it is a Fredholm map of index zero with no critical points?

$\endgroup$
3
  • 1
    $\begingroup$ No critical point + index $0$ means that the differential is an isomorphism, right? Then the inverse function theorem holds in normed vector spaces, see e.g. Lang's Introduction to Differentiable Manifolds. $\endgroup$
    – abx
    Commented Aug 7, 2019 at 8:13
  • $\begingroup$ thanks, I dont really get it why it is an isomorphism can you please explain more. $\endgroup$ Commented Aug 7, 2019 at 8:19
  • $\begingroup$ No theorem needed. Just check the definition of local diffeomorphism and of Fredholm map. $\endgroup$ Commented Aug 7, 2019 at 9:27

1 Answer 1

5
$\begingroup$

If a map is a local diffeomorphism, then it is differentiable and has a differentiable inverse, so by the chain rule the derivative is a linear isomorphism of topological vector spaces, and therefore has 0 kernel and 0 cokernel, and is therefore Fredholm with index zero. This answers the question in the title. On the other hand, if the differential at a point is Fredholm, it has finite dimensional kernel and cokernel, but those might both be positive in dimension, and any such linear map is a differentiable map. So for example a 2x2 matrix with 1 dimensional kernel and 1 dimensional cokernel has Fredholm index zero but is not a local diffeomorphism. If in addition there are no critical points, i.e. the differential is everywhere surjective, then the cokernel is zero, so if the index is zero, then the kernel is also zero, and so the differential is a linear isomorphism. If the differential is Fredholm, it is also bounded, with bounded inverse modulo compact operators, but therefore bounded inverse (by the bounded inverse theorem), so an isomorphism of topological vector spaces. We can then apply the inverse function theorem as proven by Lang.

$\endgroup$
1
  • $\begingroup$ Thanks Ben, I understand it now. $\endgroup$ Commented Aug 7, 2019 at 8:44

Not the answer you're looking for? Browse other questions tagged .